Occasional OpenGL errors under Windows OS

[leviathan117]

I’ve been attempting to create a small 3D graphics library for OpenGL. I’ve been working under Linux OS for some time but then had to move to Windows, where I’ve met a few problems. One of them is that I get runtime errors for glBindTextures function. However, absolutely the same code works correctly under Linux.

General info:
Language: Python 3.7.4
OS: Windows 10 (v10.0.18363.535)
Python interpreter: Anaconda
OpenGL version: 4.6

This is the code sample where the error happens:

    glUniform1i(glGetUniformLocation(main_context.shader_texture.get_shader(), "tex_sampler"), 0)
    for k in range(len(main_context.lights)):
        glUniform1i(glGetUniformLocation(main_context.shader_texture.get_shader(),
                                         "shadowMap[" + str(k) + "]"), k + 1)
        glBindTextures(k + 1, k + 2, main_context.lights[k].depth_map)

“main_context.shader_texture” shader code (fragment shader part) (only part for textures is shown):

    const int MAX_LIGHTS = 10;

    in vec2 textures;
    in vec3 fragment_normal;
    in vec3 fragment_pos;
    in vec4 FragPosLightSpace[MAX_LIGHTS];

    out vec4 color;
    uniform sampler2D tex_sampler;
    uniform sampler2D shadowMap[MAX_LIGHTS];

Code explanation:
I’m attempting to implement shadows. For this, I use shadow maps, which are depth textures. In this code, I’m trying to pass created depth (shadow) maps to the shader before drawing objects. Because a user may have multiple light sources, I use an array of input textures. I index them using glUniform1i, and them one by one fill using ‘glBindTextures’.

Problem:
This code crashes with this error output:

Traceback (most recent call last):
  File "C:\Users\artem\Anaconda3\lib\site-packages\OpenGL\latebind.py", line 41, in __call__
    return self._finalCall( *args, **named )
TypeError: 'NoneType' object is not callable

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "\Desktop/TRON/test_code.py", line 21, in <module>
    bla.main_loop()
  File "\Desktop\TRON\TRON.py", line 976, in main_loop
    window.draw()
  File "\Desktop\TRON\TRON.py", line 945, in draw
    i.real_draw()
  File "\Desktop\TRON\TRON.py", line 676, in real_draw
    glBindTextures(k + 1, k + 2, main_context.lights[k].depth_map)
  File "\Anaconda3\lib\site-packages\OpenGL\latebind.py", line 45, in __call__
    return self._finalCall( *args, **named )
  File "\Anaconda3\lib\site-packages\OpenGL\wrapper.py", line 677, in wrapperCall
    pyArgs = tuple( calculate_pyArgs( args ))
  File "\Anaconda3\lib\site-packages\OpenGL\wrapper.py", line 438, in calculate_pyArgs
    yield converter(args[index], self, args)
  File "\Anaconda3\lib\site-packages\OpenGL\converters.py", line 135, in __call__
    return self.function( incoming )
  File "\Anaconda3\lib\site-packages\OpenGL\arrays\arraydatatype.py", line 149, in asArray
    return cls.getHandler(value).asArray( value, typeCode or cls.typeConstant )
  File "\Anaconda3\lib\site-packages\OpenGL\arrays\arraydatatype.py", line 53, in __call__
    typ.__module__, type.__name__, repr(value)[:50]
TypeError: ('No array-type handler for type numpy.type (value: 6) registered', <OpenGL.converters.CallFuncPyConverter object at 0x000001717E1BAC08>)

Note:
Some code, like shader binding, is not shown.
File locations are shortened

[GClements]

@leviathan117

glBindTextures(k + 1, k + 2, main_context.lights[k].depth_map)

This doesn’t make sense. The second argument should probably be 1, and the third argument should be an array (the wrapper code might convert scalars to single-element arrays automatically, but it’s better not to rely upon it).

Also: have you checked that all of the main_context.lights[k].depth_map values are valid?

[leviathan117]

I’ve checked the docs for glBindTextures - and… Well, yes - my code really didn’t make any sense. However, for some reason, it worked under Linux. And I think I know why.

If we take for granted, that glBindTextures, while looping just used one texture that I passed (main_context.lights[k].depth_map) for each iteration, then this probably happened:

This is how the array of textures looked after each call of glBindTextures:

-1 -1 -1 -1 -1 -1 -1 -1 -1 -1
 1  1 -1 -1 -1 -1 -1 -1 -1 -1
 1  2  2  2 -1 -1 -1 -1 -1 -1
 1  2  3  3  3 -1 -1 -1 -1 -1

-1  - no texture specified
1 - first texture (shadow map for first light source)
2 - second texture
...

(example given for three light sources)

So some of the unused texture positions in the array were also filled. However, because I was using only up to 3 lights in my tests, the glBindTextures function never attempted to address an element outside the texture array, so everything worked ‘correctly’.

As for Windows -

Your suggested solution for rewriting the binding call as:
glBindTextures(k + 1, 1, main_context.lights[k].depth_map)
Gave out the same error.

And, yes, I’ve checked texture maps - they are correct.

[leviathan117]

Found solution:

Yes, the third argument had to be an actual array.

I rewrote the code to the following:

    glUniform1i(glGetUniformLocation(main_context.shader_texture.get_shader(), "tex_sampler"), 0)
    for k in range(len(main_context.lights)):
        glUniform1i(glGetUniformLocation(main_context.shader_texture.get_shader(),
                                         "shadowMap[" + str(k) + "]"), k + 1)
    textures_array = []
    for i in main_context.lights:
        textures_array.append(i.depth_map)
    glBindTextures(1, len(main_context.lights), textures_array)

And everything worked fine. Thank you so much for your suggestions.

As I wrote before, it seems that under Linux one texture that I passed was just duplicated many times to form an array. The same didn’t happen under Windows OS.